MS THESIS PRESENTATION
Image Super-Resolution Using Deep Feedforward Neural Networks In Spectral Domain
(Supervisor: Assoc. Prof. Dr. Selim Aksoy)
Computer Engineering Department
With recent advances in deep learning area, learning machinery and mainstream approaches in computer vision research have changed dramatically from hardcoded features combined with classifiers to end-to-end trained deep convolutional neural networks (CNN) which give the state-of-the-art results in most of the computer vision research areas. Single-image super-resolution is one of these areas which are considerably influenced by deep learning advancements. Most of the current state-of-the-art methods on super-resolution problem learn a nonlinear mapping from low-resolution images to high-resolution images in the spatial domain using consecutive convolutional layers in their network architectures. However, these state-of-the-art results are obtained by training a separate neural network architecture for each different scale factor. We propose a novel singleimage super-resolution system with the limited number of learning parameters in spectral domain in order to eliminate the necessity to train a separate neural network for each scale factor. As a spectral transform function which converts images from the spatial domain to the frequency domain, discrete cosine transform (DCT) which is a variant of discrete Fourier transform (DFT) is used. In addition, in the post-processing step, an artifact reduction module is added for removing ringing artifacts occurred due to spectral transformations. Even if the PSNR measurement of our super-resolution system is lower than current stateof-the-art methods, the spectral domain allows us to develop a single model with a single dataset for any scale factor and relatively obtain better SSIM results.
Keywords: super-resolution, deep learning, transfer learning, fourier transform.
DATE: 26 March 2018, Monday @ 09:00